JP2013-250663A discloses a driving scene recognition device. In this driving scene recognition device, information such as an operation by a driver and a behavior of a vehicle are detected and differentiation value information corresponding to differentiation values of the information are then detected. Further, based on the differentiation value information in addition to the information on the operation by the driver and the behavior of the vehicle, a symbol string corresponding to a state of the vehicle is generated, and a driving scene is recognized based on an order of driving words obtained by segmenting the symbol string. In addition, the driving scene recognition device disclosed in JP2013-250663A recognizes, for example, driving scenes in which the driver is about to start a lane change and the driver is about to park the vehicle based on the detected driver's driving operation and vehicle behavior. Further, in these recognized driving scenes, the driving is assisted by automatically displaying an image of the situation rearward of the destination lane or automatically turning on a parking assist system.
However, the driving scene recognition device disclosed in JP2013-250663A cannot estimate a pause (or boundary) of those driving scenes with sufficient accuracy, thereby causing a problem that the information cannot be fed to the driver at a suitable timing. That is, there is a large amount of information to be fed to the driver who is driving the vehicle, and if all of the information is displayed at once, the driver has to selectively recognize information that is necessary for the current driving situation out of the currently displayed information. Since the work in which the driver selectively recognizes the necessary information out of the large amount of information is very complicated and it may confuse the driver, the driver cannot fully utilize the information even if the information is illustrated.